The five questions on ethics in artificial intelligence (AI).

Ethics is defined as the moral principles governing the behavior or actions of an individual or a group. When we invent a technology, it must function ethically. Logically, it will follow the moral principles and standards of its creator. Therefore, if we are mindful of our ethical and moral obligations to society, these concerns will find their way into the functioning, and use, of the technology, for example, artificial intelligence. (THE FINTECH TIMES)

How should we ensure that we are ethical in our use of artificial intelligence? According to Andrew Brockway, CTO, Confused.com, there are five questions to ask ourselves.

How will AI benefit those it is meant to serve?

If the objective of the technology is to compare a financial product, does it produce authentic answers that will really help consumers save money? If the AI algo helps find consumers a better deal, it is doing good for society and is an ethical success on this parameter.

Ethical AI: Will the cost of AI’s disruption of society exceed its benefits?

In other words, how the technology integrates with humankind in society. Does it merely replace people’s jobs, or does it free them up to do more meaningful and creative work?

“From a moral standing, most people would agree that the ability to reduce costs, enhance quality and free up the time of frontline staff to do work that is beneficial to all of society is a positive thing,” says Brockway.

Is the data free of bias?

There is an old saying in technology: “Garbage in Garbage out.”

So it is in AI – it is only as good as the data behind it. If the data on which the algo was trained is biased, results from running that algo will suffer from the same bias.

In 2014, Amazon (NASDAQ: AMZN) deployed the use of more than 500 algorithms to automate its process for reviewing the resumes of engineers and coders. The data used to train the algorithms was the database of resumes of the software teams, who were mostly male. When reviewing applicants’ resumes, the AI disqualified anybody from a women’s college, though that was not the intention of the creators of the algos. This issue was reported by Reuters.

Compliance with regulations

The use of the AI must comply with both external regulations and the organization’s internal charter of policies.

When we deploy AI we must test it thoroughly for potential violations of these regulations and policies.

A famous ‘redlining’ case bears repetition. In 2000, Wells Fargo (NYSE: WFC) commenced the use of a “community calculator” on its website. The tool guided home buyers to the right neighborhood, and therefore a mortgage. But the AI algo based its findings on the users’ current ZIP code and its demographics, and therefore recommended only locations with similar demographics. This introduced bias into the results with the potential for discrimination and violation of the law.

Ethical AI: Should we deploy AI just because it is fashionable?

“Benefitting stakeholders, customers, and wider society should be the main driving force behind any decision to employ AI or not,” observes Brockway. “But equally, we need to view the capabilities of the application scrupulously, highlighting the limitations of the data in order to address the shortcomings. Just because AI allows you to do something, it doesn’t mean you should.”

Related Story: Artificial Intelligence: The Pentagon Adopts Five AI Principles